Quantitative results

Measures of the activity, safety and quality in clinical processes generally improved. Table 2 shows selected examples. Waiting times and length of stay were reduced while the readmission rate was stable. Measures of patients’ safety and outcomes also generally improved. Overall, 30-day mortality and the incidence of hospital-acquired infections declined, and survival after treatments for cancers and cardiovascular disease increased. These results, could, however, not be continuously assessed because national clinical quality registries published results annually. Therefore, attention both within the organisation and in the media drifted towards the process measures, and the changes tended to be perceived as efficiency only, and not quality improvement.

Table 2 also shows that the budget deficits were eliminated over 2 years followed by accounting balance or surpluses during the next 7 years. Cost-effectiveness, measured as the number of DRG points per full-time equivalent (FTE), improved. The main mechanism was that the previous continuous increase in the number of FTEs levelled off when units merged and fewer bed wards operated continuously. The financial equity balances improved, and after 5 years loans were granted. Long-awaited investments in renewal of medical equipment and information technology (eg, a joint electronic health record for our hospitals) were expedited. We planned and built a new 230-bed patient hotel, necessary to support the shift from admissions to outpatient treatments in a region with long travel distances, a positron emission tomography centre with a cyclotron, crucial to our function as the regions’ cancer centre, and a new hospital wing with modern facilities for day surgery, laboratories and intensive care, during my 10-year tenure.

Measures of patient satisfaction available from national authorities’ annual surveys showed stable or slightly (in some domains) improved results. In contrast, corresponding surveys of employees showed decreasing job satisfaction. Staff reported less autonomy, increased workload, more bureaucracy, inefficient internal communication, and measurement and improvement fatigue. A perception of reduced freedom to speak out was increasingly voiced, especially from the physicians’ union.

First lesson learnt: data-driven improvement works, but it comes at a cost

It is well established that management using balanced scorecards can drive performance.6 This is in accordance with my experience, as I consider quantified goals and continuous measurements a central mechanism behind the change and improvement achieved. The implementation of KPIs and an electronic LIS combined with performance agreement contracts for leaders rapidly changed the organisation’s focus.

I was, however, not sufficiently aware of the risks. Composition of a balanced scorecard of KPIs relevant across organisational levels and units was difficult to negotiate. Professionals struggled to agree on the selection, and few were satisfied with the compromises reached. Imbalance in data availability, especially between process and outcome measures, skewed attention away from the core of the strategy. Also, central authorities superimposed new goals and KPIs on those prioritised by us, and middle managers faced conflicting goals.

I was familiar with data-driven improvement from clinical research within neurosurgery, but not sufficiently aware of the contradictions and pitfalls of performance management in complex organisations. As a consequence, I was surprised when the organisation I thought I knew paradoxically reported reduced job satisfaction, while most KPIs showed improvements. This necessitated a comprehensive, long-lasting and resource-demanding communication effort with middle managers, union representatives and employees. Recently, scholars addressed the high price of efficiency, and suggested organisations should offer more attention to resilience.7 8 This could make sense. On the other hand, building in the recommended slack in a public hospital which is restrained by not only financial resources, but also staff shortages, is difficult.

Second lesson learnt: it is inherently difficult to recognise when and how an issue changes from a professional to a political logic

Within the UNN, we presupposed the study of how PCI should be organised would follow the usual professional logic. We expected that registry data about patients’ outcomes and the absence of geographical variation would be decisive, and did not sense what was under discussion among politicians and in local media in the southern part of the region. We were taken by surprise when 18 mayors campaigned for the decentralisation, arguing that one of them had almost died from an acute MI during air ambulance transfer to the UNN. The groundless contention that patients residing in the southern part of the region had inferior outcomes after acute MI had become ‘common knowledge’ before we became aware of the situation. Politicians in our part of the region slowly mobilised to preserve what they perceived to be a high-quality service, only to escalate the conflict.

Geographically based conflicting interests should be of no surprise to healthcare leaders. I knew this, but nonetheless learnt that it is inherently difficult to recognise when and how the discourse of a process changes from a professional to a political logic. In hindsight, the board could have used their political network better to retrieve information at an earlier stage, and as the tertiary referral hospital for the entire region, we should obviously have surveilled all local media more rigorously.

Third lesson learnt: during conflict, role clarity is crucial

The head of UNN board and I agreed that I should communicate directly with the RHA and its board, since we perceived the issue to be mainly of medical professional character. This was in accordance with previous practice, but unruly in the specific situation, as it is the head of the enterprise board (and not the CEO) who reports to the board at the superior regional level. This lack of role clarity shifted attention away from the issue under debate to procedural formalities. This weakened our case and caused the abrupt breakdown of trust and change of leadership.

Fourth lesson learnt: be prepared to resign when strategically unaligned with superior authorities

In hindsight, it gradually became clear over a period of months that a disagreement about strategic direction had developed between the UNN and the RHA. I was not sufficiently conscious of my own attitude, and in consequence, not transparent regarding my intentions. I hoped they would bend, and return to a professional logic.

I never made it clear that I was not willing to lead what I perceived as dismantling of a core function in the university hospital. I was afraid that stating this would be understood as an inappropriate ‘threat to resign’. Research shows, however, that hospitals whose leaders are strategically unaligned with superior authorities, stagnate.10 Therefore, it would have been frank to make my position clear. It complicated matters that our board, who I reported to, followed the same path. I should, in any case, have understood that a board strategically unaligned with the superior authority who appoints it, would be replaced.

Paradoxically, our organisation came together during this process. Both I as the CEO and the board experienced unconditional support. The internal tensions described above faded, and employees applauded a leadership that stood up for the organisation’s values and strategy. This is of course a well-known phenomenon when facing a ‘common enemy’. It strengthened my aftermath, but obviously caused a weakening of the confidence and collaboration in the region.

Fifth lesson learnt: 10 years is enough!

As mentioned, I was considering resigning when the situation that prompted my resignation emerged. Evidence from studies of commercial industries shows an inverted U-shaped relationship between CEO tenure and profitability,11 and studies of hospitals indicate that frequent CEO churns impact quality systems negatively.12 Taken together, continuity in leadership is favourable, but the risk of stagnation is high after 10 years. It is probably time to step down for a CEO once one has started considering it. I think I inexpediently postponed the decision because I wanted to experience some of the milestones to be reached, such as celebrating the opening of new facilities.

Finally, should I have accepted the CEO challenge in the first place? By then, I believed it was important that hospitals were led by doctors. Recent studies report an association between physician CEOs, better treatment outcomes and higher patient satisfaction.13 14 Cross-sectional comparisons are, however, prone to selection bias: it is difficult to know whether good hospitals recruit physician CEOs easier, or whether such CEOs actually change their hospitals for the better.